Componentwise localization of solutions to systems of operator inclusions via Harnack type inequalities

2 years ago

Abstract

We establish compression-expansion type fixed point theorems for systems of operator inclusions with decomposable multivalued maps. The approach is vectorial allowing to localize individually the components of solutions and to obtain multiple solutions with multiplicity not necessarily concerned with all components of the solution. A general scheme of applicability of the theory is elaborated based on Harnack type inequalities and illustrated on systems of diﬀerential inclusions with one-dimentional ϕ-Laplacian.

Authors

Radu Precup
Department of Mathematics Babes-Bolyai University, Cluj-Napoca, Romania

Jorge Rodríguez-López
Departamento de Estatística, Análise Matemática e Optimización, CITMAga, Universidade de Santiago de Compostela, 15782, Facultade de Matemáticas, Campus Vida, Santiago, Spain.

Keywords

Compression-expansion fixed point theorem; Harnack type inequality; positive solution; operator inclusion; nonlinear system; ϕ-Laplacian

Paper coordinates

R. Precup, J. Rodríguez-López, Componentwise localization of solutions to systems of operator inclusions via Harnack type inequalities, Quaestiones Mathematicae, 46 (2023) no. 7, pp. 1481-1496, https://doi.org/10.2989/16073606.2022.2107959

PDF

https://www.tandfonline.com/doi/abs/10.2989/16073606.2022.2107959

About this paper

Journal

Quaestiones Mathematicae

Publisher Name

Taylor and Francis

DOI

https://doi.org/10.2989/16073606.2022.2107959

Print ISSN

1607-3606

Online ISSN

1727-933X

google scholar link

Paper (preprint) in HTML form

Componentwise localization of solutions to systems of operator inclusions via Harnack type inequalities

Radu Precup Department of Mathematics, Babeş-Bolyai University, 400084 Cluj-Napoca, Romania, E-mail: r.precup@math.ubbcluj.ro and Jorge Rodríguez-López Departamento de Estatística, Análise Matemática e Optimización, Instituto de Matemáticas, Universidade de Santiago de Compostela, 15782, Facultade de Matemáticas, Campus Vida, Santiago, Spain, E-mail: jorgerodriguez.lopez@usc.es

Abstract.

We establish compression-expansion type fixed point theorems for systems of operator inclusions with decomposable multivalued maps. The approach is vectorial allowing to localize individually the components of solutions and to obtain multiple solutions with multiplicity not necessarily concerned with all components of the solution. A general scheme of applicability of the theory is elaborated based on Harnack type inequalities and illustrated on systems of differential inclusions with one-dimentional $\phi$ -Laplacian.

Mathematics Subject Classification: 47H10; 34A60; 34B18

Keywords and phrases: Compression–expansion fixed point theorem, Harnack type inequality, positive solution, operator inclusion, nonlinear system, $\phi$ -Laplacian.

1. Introduction

The main goal of this paper is to discuss the existence, componentwise localization and multiplicity of positive solutions for systems of differential inclusions with one-dimensional $\phi$ -Laplacian, namely

(1.1)

\left\{\begin{array}[]{l}-\left(\phi_{1}(u_{1}^{\prime})\right)^{\prime}\in G_% {1}(t,u_{1},u_{2})\\ -\left(\phi_{2}(u_{2}^{\prime})\right)^{\prime}\in G_{2}(t,u_{1},u_{2}),\end{% array}\right.

for $t\in\left(0,1\right),$ subject to the boundary conditions

u_{1}^{\prime}\left(0\right)=u_{2}^{\prime}\left(0\right)=u_{1}\left(1\right)=% u_{2}\left(1\right)=0,

where for each $i\in\{1,2\},$ $\phi_{i}:(-a_{i},a_{i})\rightarrow(-b_{i},b_{i})$ $\ (0<a_{i},\ b_{i}\leq+\infty)$ is an increasing homeomorphism with $\phi_{i}(0)=0$ and $G_{i}$ is a multivalued map. In particular, the homeomorphisms $\phi_{1},\phi_{2}$ can be one of the following

\phi:\mathbb{R}\rightarrow\mathbb{R},\ \ \ \phi\left(x\right)=\left|x\right|^{% p-2}x;

\phi:\mathbb{R}\rightarrow\left(-1,1\right),\ \ \ \phi\left(x\right)=\frac{x}{% \sqrt{1+x^{2}}};

\phi:\left(-1,1\right)\rightarrow\mathbb{R},\ \ \ \phi\left(x\right)=\frac{x}{% \sqrt{1-x^{2}}},

corresponding to the one-dimensional $p$ -Laplace operator, the mean curvature operator in the Euclidian and Minkowski space, respectively. Such type of equations involving the $\phi$ -Laplacian has been investigated in a large number of papers using fixed point methods, degree theory, upper and lower solution techniques and variational methods. We refer the reader to the papers [1, 2, 3, 4, 5, 6, 7, 8, 10, 11, 12, 14, 15, 17, 18, 24], to the survey work [16], and the bibliographies therein.

It is worth noting that in our work on system (1.1), the two homeomorphisms $\phi_{1},\phi_{2}$ can be of different types. Also, for a solution $u=\left(u_{1},u_{2}\right),$ different localizations of the components $u_{1}$ and $u_{2}$ are possible, allowing to obtain multiple solution results where the multiplicity concerns either only one of the components or both of them.

More general, we shall discuss systems of operator inclusions in Banach spaces, of the form

(1.2)

\left\{\begin{array}[]{l}L_{1}\left(u_{1}\right)\in F_{1}\left(u_{1},u_{2}% \right)\\ L_{2}\left(u_{2}\right)\in F_{2}\left(u_{1},u_{2}\right),\end{array}\right.

where, in this abstract setting, the ‘boundary conditions’ are simulated by the belonging of $u_{i}$ to the domain $D\left(L_{i}\right)$ of the not necessarily linear operator $L_{i}$ $\left(i=1,2\right),$ while the ‘positivity’ of solutions means their belonging to a given cone.

If the operators $L_{i}\ \left(i=1,2\right)$ are invertible, as is the case of the boundary value problem related to system (1.1), then (1.2) is equivalent to the fixed point problem

(1.3)

\left\{\begin{array}[]{c}u_{1}\in L_{1}^{-1}F_{1}(u_{1},u_{2})\\ u_{2}\in L_{2}^{-1}F_{2}(u_{1},u_{2}),\end{array}\right.

with decomposable multivalued maps $L_{1}^{-1}F_{1}$ and $L_{2}^{-1}F_{2}.$

Thus, it is natural to develop a theory for abstract systems of inclusions, of the form

(1.4)

\left\{\begin{array}[]{l}u_{1}\in N_{1}(u_{1},u_{2})\\ u_{2}\in N_{2}(u_{1},u_{2}),\end{array}\right.

in a Banach space $X,$ where $N=(N_{1},N_{2}):A\subset X^{2}\rightarrow 2^{X^{2}}$ is a multivalued operator such that each of its components is a decomposable map, that is, the composition of two multivalued maps

N_{i}(u)=\Psi_{i}(u,\Phi_{i}(u)),\quad u=\left(u_{1},u_{2}\right)\in A\qquad(i% =1,2).

Our theoretical strategy is to make the reverse path, from abstract to concrete. Thus, first, in Section 2, we develop a fixed point theory with localization on components for the general system (1.4); next, in Section 3, we apply this theory to systems of type (1.2); and, finally, in Section 4, we particularize even more to obtain results for the considered boundary value problem related to system (1.1). Of course, one may use the abstract theory to obtain similar results for system (1.1) subject to some other boundary conditions. Also the results can be immediately extended to systems of more than two inclusions. Our restriction to systems of two inclusions is only to simplify the presentation.

The common approach to systems is to look at them as generalizations of equations. On the contrary, in our vectorial approch, a system is seen as a particular equation that has the splitting property. Consequently, a larger diversity of results may be expected in the case of systems.

This paper extends for inclusions the theory established in [9], [20], [21] and [22] for equations, and in [13] for inclusions with convex-valued maps. The extension is not trivial since, as explain in [19], several difficulties arise when treating compositions of multivalued maps. One of them consists in guaranteeing continuity properties for the maps, another one concerns the geometric properties of their values. For example, even if one of the maps is single-valued (but nonlinear) and the values of the other one are convex, the values of the composed map can be non-convex, contrary to the hypotheses of basic fixed point theorems for multivalued maps.

2. Existence results for operator systems with decomposable multivalued maps

Let $X$ be a Banach space and consider the fixed point problem

\left\{\begin{array}[]{l}u_{1}\in N_{1}(u_{1},u_{2})\\ u_{2}\in N_{2}(u_{1},u_{2}),\end{array}\right.

where $N=(N_{1},N_{2}):A\subset X^{2}\rightarrow 2^{A}$ is a multivalued operator such that each of its components is a decomposable map, that is, the composition of two upper semicontinuous (usc, for short) multivalued maps:

N_{i}(u)=\Psi_{i}(u,\Phi_{i}(u)),\quad u=\left(u_{1},u_{2}\right)\in A\qquad(i% =1,2).

2.1. A Schauder type fixed point theorem

First, we present a generalization of Kakutani’s fixed point theorem for a system of two decomposable maps. For the case of a single equation, the reader may find a similar result in [19].

Theorem 2.1.

Let $A$ be a nonempty, convex and compact subset of $X^{2}$ , $Y_{1}$ , $Y_{2}$ Banach spaces and $N=(N_{1},N_{2}):A\subset X^{2}\rightarrow 2^{A}$ , $(N_{1},N_{2})(u)=(\Psi_{1}(u,\Phi_{1}(u)),\Psi_{2}(u,\Phi_{2}(u)))$ , a multivalued operator such that for each $i\in\{1,2\}$ ,

(i)

$\Phi_{i}:A\rightarrow 2^{Y_{i}}$ is usc with compact convex values;
(ii)

$\Psi_{i}:A\times Y_{i}\rightarrow 2^{X}$ is usc with closed convex values.

Then $N$ has a fixed point in $A$ .

Proof.

Consider the set $B=A\times\overline{\mathrm{co}}\Phi_{1}(A)\times\overline{\mathrm{co}}\Phi_{2}% (A)$ . Since $A$ is compact and $\Phi_{1}$ is usc with compact values, $\overline{\mathrm{co}}\Phi_{1}(A)$ is compact (similarly, $\overline{\mathrm{co}}\Phi_{2}(A)$ is also compact). Hence, $B$ is the Cartesian product of compact convex sets, so it is a compact convex subset of $X^{2}\times Y^{2}$ .

Now, we associate to $N$ the map $\Pi:B\rightarrow 2^{B}$ given by

\Pi(u_{1},u_{2},v_{1},v_{2})=\Psi_{1}(u_{1},u_{2},v_{1})\times\Psi_{2}(u_{1},u% _{2},v_{2})\times\Phi_{1}(u_{1},u_{2})\times\Phi_{2}(u_{1},u_{2}).

Note that $\Pi$ is usc with closed convex values and thus the Bohnenblust–Karlin fixed point theorem ensures that $\Pi$ has a fixed point $(u_{1},u_{2},v_{1},v_{2})\in B$ , that is,

u_{1}\in\Psi_{1}(u_{1},u_{2},v_{1}),\quad u_{2}\in\Psi_{2}(u_{1},u_{2},v_{2}),% \quad v_{1}\in\Phi_{1}(u_{1},u_{2}),\quad v_{2}\in\Phi_{2}(u_{1},u_{2}).

Therefore, the pair $(u_{1},u_{2})\in A$ is a fixed point of $N$ . ∎

Corollary 2.2.

Let $C$ be a nonempty, convex and closed subset of $X^{2}$ , $Y_{1}$ , $Y_{2}$ Banach spaces and $N=(N_{1},N_{2}):C\subset X^{2}\rightarrow 2^{C}$ , $(N_{1},N_{2})(u)=(\Psi_{1}(u,\Phi_{1}(u)),\Psi_{2}(u,\Phi_{2}(u)))$ , a multivalued operator such that $N(C)$ is relatively compact and for each $i\in\{1,2\}$ ,

(i)

$\Phi_{i}:C\rightarrow 2^{Y_{i}}$ is usc with compact convex values;
(ii)

$\Psi_{i}:C\times Y_{i}\rightarrow 2^{X}$ is usc with closed convex values.

Then $N$ has a fixed point in $C$ .

Proof.

It suffices to take the compact and convex set $A=\overline{\mathrm{co}}N(C)\subset C$ and to apply Theorem 2.1. ∎

2.2. A vector version of Krasnosel’skiĭ’s fixed point theorem in cones

Now we present the vector version of Krasnosel’skiĭ’s fixed point theorem in cones for the class of decomposable maps. Note that this type of componentwise localization of positive fixed points for systems was initiated in [20, 21] for single-valued compact maps and continued in [13] by a similar result for usc multivalued maps. Also, a Krasnosel’skiĭ type fixed point theorem in cones for a single decomposable map was proved in [5] (see also [23] for a fixed point index theory for decomposable maps).

Recall that a closed convex subset $K$ of a Banach space $X$ is a cone if it satisfies that $K\cap(-K)=\{0\}$ and $\lambda x\in K$ for every $x\in K$ and for all $\lambda\geq 0$ . Any cone $K$ induces a partial order relation in $X$ , denoted by $\preceq$ , namely $x\preceq y$ if $y-x\in K$ . One says that $x\prec y$ if $y-x\in K\setminus\{0\}$ .

In the sequel, let $K_{1}$ and $K_{2}$ be two cones of the Banach space $X$ and so $K:=K_{1}\times K_{2}$ is a cone of $X^{2}$ . For $r,R\in\mathbb{R}_{+}^{2}$ , $r=(r_{1},r_{2})$ , $R=(R_{1},R_{2})$ , we denote

	$\displaystyle(K_{i})_{r_{i},R_{i}}$	$\displaystyle:=\{u\in K_{i}\,:\,r_{i}\leq\left\\|u\right\\|\leq R_{i}\}\quad(i=1% ,2),$
	$\displaystyle K_{r,R}$	$\displaystyle:=\{u=\left(u_{1},u_{2}\right)\in K\,:\,r_{i}\leq\left\\|u_{i}% \right\\|\leq R_{i}\quad\text{for }i=1,2\}.$

We look for fixed points of an operator $N=(N_{1},N_{2}):K\longrightarrow 2^{K}$ , whose components $N_{1}$ and $N_{2}$ are of the form

N_{i}(u)=\Psi_{i}(u,\Phi_{i}(u)),\quad u\in K\quad(i=1,2).

Here we assume that for a Banach space $Y$ and two closed convex sets $K_{Y_{1}},K_{Y_{2}}\subset Y,$ the following conditions on $\Psi_{i}$ and $\Phi_{i}$ are satisfied for each $i\in\{1,2\}$ :

(H_Φ)

$\Phi_{i}:K\rightarrow 2^{K_{Y_{i}}}$ is usc with compact convex values;
(H_Ψ)

$\Psi_{i}:K\times K_{Y_{i}}\rightarrow 2^{K_{i}}$ is usc with closed convex values.

Now we state and prove the main result of this section.

Theorem 2.3.

Let $\alpha_{i},\beta_{i}>0$ with $\alpha_{i}\neq\beta_{i}$ , $r_{i}=\min\{\alpha_{i},\beta_{i}\}$ and $R_{i}=\max\{\alpha_{i},\beta_{i}\}$ for $i=1,2$ . Assume that $N:K_{r,R}\rightarrow 2^{K}$ ,

N(u)=(\Psi_{1}(u,\Phi_{1}(u)),\Psi_{2}(u,\Phi_{2}(u))),

$N(K_{r,R})$ is relatively compact and for each $i\in\{1,2\},$ $\Phi_{i}$ and $\Psi_{i}$ satisfy (H_Φ) and (H_Ψ), respectively. In addition assume that the following conditions

	$\displaystyle u_{i}$	$\displaystyle\not\in N_{i}(u_{1},u_{2})+\mu h_{i}\quad\text{ for any }(u_{1},u% _{2})\in K_{r,R}\text{ with }\left\\|u_{i}\right\\|=\alpha_{i}\text{ and }\mu>0;$
	$\displaystyle\lambda u_{i}$	$\displaystyle\not\in N_{i}(u_{1},u_{2})\quad\text{ for any }(u_{1},u_{2})\in K% _{r,R}\text{ with }\left\\|u_{i}\right\\|=\beta_{i}\text{ and }\lambda>1$

hold for some $h_{i}\in K_{i}\setminus\{0\}\ \left(i=1,2\right).$ Then $N$ has a fixed point $u=(u_{1},u_{2})\in K$ with $r_{i}\leq\left\|u_{i}\right\|\leq R_{i}$ for $i=1,2$ .

Proof.

We consider four cases which cover all the possible combinations of compression–expansion conditions for $N_{1}$ and $N_{2}$ .

Case 1: Assume first that $\alpha_{i}<\beta_{i}$ for both $i=1,2$ (compression for both $N_{1}$ and $N_{2}$ ). Then $r_{i}=\alpha_{i}$ and $R_{i}=\beta_{i}$ for $i=1,2$ . Define the map $\tilde{N}=(\tilde{N}_{1},\tilde{N}_{2}):K\rightarrow 2^{K}$ by

\tilde{N}_{i}(u_{1},u_{2})=\mu(u_{1},u_{2})N_{i}\left(\delta_{1}(u_{1})\frac{u% _{1}}{\left\|u_{1}\right\|},\delta_{2}(u_{2})\frac{u_{2}}{\left\|u_{2}\right\|% }\right)+\left(1-\mu(u_{1},u_{2})\right)h_{i},

where $\mu(u_{1},u_{2})=\min\left\{\left\|u_{1}\right\|/r_{1},\left\|u_{2}\right\|/r_% {2},1\right\}$ and $\delta_{i}(u_{i})=\max\{\min\{u_{i},R_{i}\},r_{i}\}$ for $i=1,2$ . The map $\tilde{N}$ is a decomposable map, namely, $\tilde{N}_{i}(u_{1},u_{2})=\tilde{\Psi}_{i}(u_{1},u_{2},\tilde{\Phi}_{i}(u_{1}% ,u_{2}))$ ( $i=1,2$ ), with

	$\displaystyle\tilde{\Phi}_{i}(u_{1},u_{2})$	$\displaystyle=\Phi_{i}\left(\delta_{1}(u_{1})\frac{u_{1}}{\left\\|u_{1}\right\\|% },\delta_{2}(u_{2})\frac{u_{2}}{\left\\|u_{2}\right\\|}\right),$
	$\displaystyle\tilde{\Psi}_{i}(u_{1},u_{2},v)$	$\displaystyle=\mu(u_{1},u_{2})\Psi_{i}\left(\delta_{1}(u_{1})\frac{u_{1}}{% \left\\|u_{1}\right\\|},\delta_{2}(u_{2})\frac{u_{2}}{\left\\|u_{2}\right\\|},v% \right)+\left(1-\mu(u_{1},u_{2})\right)h_{i}.$

Observe that $\tilde{\Phi}_{i}$ and $\tilde{\Psi}_{i}$ satisfy conditions (H_Φ) and (H_Ψ), respectively. Moreover, note that $\tilde{N}(K)$ is relatively compact since its values belong to the compact set

C=\overline{\mathrm{co}}\left(N(K_{r,R})\cup\{h\}\right),

where $h=\left(h_{1},h_{2}\right).$ Therefore, Corollary 2.2 applies and guarantees the existence of a fixed point of $\tilde{N}$ in $C\subset K$ . It can be shown in a similar way to the proof of [13, Theorem 2] (or [20, 21]) that the fixed point $u$ belongs to $K_{r,R}$ . Hence, $u$ is also a fixed point of the operator $N$ which is located in $K_{r,R}$ .

Case 2: Assume that $\alpha_{1}<\beta_{1}$ (compression for $N_{1}$ ) and $\beta_{2}<\alpha_{2}$ (expansion for $N_{2}$ ). Note that this case can be reduced to the previous one by considering the operator $N^{\ast}=(N_{1}^{\ast},N_{2}^{\ast}):K\rightarrow 2^{K}$ given by

	$\displaystyle N_{1}^{\ast}(u_{1},u_{2})$	$\displaystyle=N_{1}\left(u_{1},\left(\frac{R_{2}}{\left\\|u_{2}\right\\|}+\frac{% r_{2}}{\left\\|u_{2}\right\\|}-1\right)u_{2}\right),$
	$\displaystyle N_{2}^{\ast}(u_{1},u_{2})$	$\displaystyle=\left(\frac{R_{2}}{\left\\|u_{2}\right\\|}+\frac{r_{2}}{\left\\|u_{% 2}\right\\|}-1\right)^{-1}N_{2}\left(u_{1},\left(\frac{R_{2}}{\left\\|u_{2}% \right\\|}+\frac{r_{2}}{\left\\|u_{2}\right\\|}-1\right)u_{2}\right).$

Each component of $N^{\ast}$ can be written as the composition $N_{i}^{\ast}(u)=\Psi_{i}^{\ast}(u,\Phi_{i}^{\ast}(u))$ , where

	$\displaystyle\Phi_{i}^{\ast}(u_{1},u_{2})$	$\displaystyle=\Phi_{i}\left(u_{1},\left(\frac{R_{2}}{\left\\|u_{2}\right\\|}+% \frac{r_{2}}{\left\\|u_{2}\right\\|}-1\right)u_{2}\right)\qquad(i=1,2),$
	$\displaystyle\Psi_{1}^{\ast}(u_{1},u_{2},v)$	$\displaystyle=\Psi_{1}\left(u_{1},\left(\frac{R_{2}}{\left\\|u_{2}\right\\|}+% \frac{r_{2}}{\left\\|u_{2}\right\\|}-1\right)u_{2},v\right),$
	$\displaystyle\Psi_{2}^{\ast}(u_{1},u_{2},v)$	$\displaystyle=\left(\frac{R_{2}}{\left\\|u_{2}\right\\|}+\frac{r_{2}}{\left\\|u_{% 2}\right\\|}-1\right)^{-1}\Psi_{2}\left(u_{1},\left(\frac{R_{2}}{\left\\|u_{2}% \right\\|}+\frac{r_{2}}{\left\\|u_{2}\right\\|}-1\right)u_{2},v\right).$

Notice that the map $N^{\ast}$ is in case 1 and thus it has a fixed point $v\in K_{r,R}$ . Then the point $u=(u_{1},u_{2})$ defined as $u_{1}=v_{1}$ and $u_{2}=\left(\frac{R_{2}}{\left\|u_{2}\right\|}+\frac{r_{2}}{\left\|u_{2}\right% \|}-1\right)v_{2}$ is a fixed point of the operator $N$ .

Case 3: Assume that $\beta_{1}<\alpha_{1}$ (expansion for $N_{1}$ ) and $\alpha_{2}<\beta_{2}$ (compression for $N_{2}$ ). This case is completely analogous to the previous one.

Case 4: Assume that $\beta_{i}<\alpha_{i}$ for $i=1,2$ (expansion for both $N_{1}$ and $N_{2}$ ). This situation reduces to case 1 by taking the decomposable map $N^{\ast\ast}=(N_{1}^{\ast\ast},N_{2}^{\ast\ast})$ defined as

	$\displaystyle N_{1}^{\ast\ast}(u_{1},u_{2})$	$\displaystyle=\left(\frac{R_{1}}{\left\\|u_{1}\right\\|}+\frac{r_{1}}{\left\\|u_{% 1}\right\\|}-1\right)^{-1}N_{1}\left(\left(\frac{R_{1}}{\left\\|u_{1}\right\\|}+% \frac{r_{1}}{\left\\|u_{1}\right\\|}-1\right)u_{1},u_{2}\right),$
	$\displaystyle N_{2}^{\ast\ast}(u_{1},u_{2})$	$\displaystyle=\left(\frac{R_{2}}{\left\\|u_{2}\right\\|}+\frac{r_{2}}{\left\\|u_{% 2}\right\\|}-1\right)^{-1}N_{2}\left(u_{1},\left(\frac{R_{2}}{\left\\|u_{2}% \right\\|}+\frac{r_{2}}{\left\\|u_{2}\right\\|}-1\right)u_{2}\right).$

Thus, the proof is finished. ∎

In particular, if $\Psi(u,v)=\Psi(v)$ , then from Theorem 2.3 we obtain the following compression–expansion fixed point theorem for inclusion systems involving compositions of the form $\Psi\circ\Phi$ .

Theorem 2.4.

Let $\alpha_{i},\beta_{i}>0$ with $\alpha_{i}\neq\beta_{i}$ , $r_{i}=\min\{\alpha_{i},\beta_{i}\}$ and $R_{i}=\max\{\alpha_{i},\beta_{i}\}$ for $i=1,2$ . Assume that $N:K_{r,R}\rightarrow 2^{K}$ ,

N=(\Psi_{1}\Phi_{1},\Psi_{2}\Phi_{2}),

$N(K_{r,R})$ is relatively compact and for each $i\in\{1,2\},$ $\Phi_{i}$ and $\Psi_{i}$ satisfy (H_Φ) and (H_Ψ), respectively. In addition assume that the following conditions

	$\displaystyle u_{i}$	$\displaystyle\not\in N_{i}(u_{1},u_{2})+\mu h_{i}\quad\text{ for any }(u_{1},u% _{2})\in K_{r,R}\text{ with }\left\\|u_{i}\right\\|=\alpha_{i}\text{ and }\mu>0;$
	$\displaystyle\lambda u_{i}$	$\displaystyle\not\in N_{i}(u_{1},u_{2})\quad\text{ for any }(u_{1},u_{2})\in K% _{r,R}\text{ with }\left\\|u_{i}\right\\|=\beta_{i}\text{ and }\lambda>1$

hold for some $h_{i}\in K_{i}\setminus\{0\}\ \left(i=1,2\right).$ Then $N$ has a fixed point $u=(u_{1},u_{2})\in K$ with $r_{i}\leq\left\|u_{i}\right\|\leq R_{i}$ for $i=1,2$ .

As a straightforward consequence we obtain the following result.

Corollary 2.5.

Let $\alpha_{i},\beta_{i}>0$ with $\alpha_{i}\neq\beta_{i}$ , $r_{i}=\min\{\alpha_{i},\beta_{i}\}$ and $R_{i}=\max\{\alpha_{i},\beta_{i}\}$ for $i=1,2$ . Assume that $N:K_{r,R}\rightarrow 2^{K}$ ,

N=(\Psi_{1}\Phi_{1},\Psi_{2}\Phi_{2}),

$N(K_{r,R})$ is relatively compact and for each $i\in\{1,2\},$ $\Phi_{i}$ and $\Psi_{i}$ satisfy (H_Φ) and (H_Ψ), respectively. In addition assume that the following conditions hold:

	$\displaystyle y$	$\displaystyle\nprec u_{i}\quad\text{ for all }y\in N_{i}(u_{1},u_{2}),\ (u_{1}% ,u_{2})\in K_{r,R}\text{ with }\left\\|u_{i}\right\\|=\alpha_{i};$
	$\displaystyle y$	$\displaystyle\nsucc u_{i}\quad\text{ for all }y\in N_{i}(u_{1},u_{2}),\ (u_{1}% ,u_{2})\in K_{r,R}\text{ with }\left\\|u_{i}\right\\|=\beta_{i}.$

Then $N$ has a fixed point $u=(u_{1},u_{2})\in K$ with $r_{i}\leq\left\|u_{i}\right\|\leq R_{i}$ for $i=1,2$ .

3. Positive solutions for systems of operator inclusions

In this section, we apply the general fixed point theorems established from above to the following system of operator inclusions

(3.1)

\left\{\begin{array}[]{l}L_{i}(u_{i})\in F_{i}(u_{1},u_{2})\\ u_{i}\in D(L_{i})\quad(i=1,2),\end{array}\right.

where for each $i$ , $L_{i}:D(L_{i})\subset X\rightarrow Y$ is a not necessarily linear map which is invertible, $F_{i}:X\times X\rightarrow 2^{Y}$ is a set-valued map, and $X,\ Y$ are Banach spaces with continuous embedding $X\subset Y.$

Equivalently, we deal with the following inclusion system of type (1.4),

(3.2)

u_{i}\in L_{i}^{-1}F_{i}(u_{1},u_{2}),\quad u_{i}\in X\quad(i=1,2).

We look for positive solutions for (3.1), that is, solutions $u=(u_{1},u_{2})$ with $u_{i}\in K_{0}^{i}\cap X$ , where $K_{0}^{i}\subset Y$ is a cone for every $i=1,2$ . We use the same symbol $\preceq$ to denote the ordering in $Y$ induced either by $K_{0}^{1}$ or $K_{0}^{2}$ . Moreover, we assume that $L_{i}^{-1}(K_{0}^{i})\subset K_{0}^{i}$ for every $i\in\{1,2\}$ .

Let $P$ be a cone in $X$ and for each $i\in\{1,2\},$ consider the following basic assumptions:

(H₁)

$L_{i}^{-1}:K_{0}^{i}\rightarrow D(L_{i})$ can be written as the composition $L_{i}^{-1}=T_{i}\circ S_{i}$ , where
1. (a)
  
  $S_{i}:K_{0}^{i}\rightarrow P$ is a continuous linear operator which maps bounded sets into relatively compact sets and $S_{i}\circ F_{i}$ has closed and convex values;
2. (b)
  
  $T_{i}:P\rightarrow D(L_{i})$ is a continuous map.
(H₂)

$S_{i}$ and $T_{i}$ are positive and increasing, i.e.,

$0\preceq u\preceq v\quad\text{implies}\quad 0\preceq S_{i}(u)\preceq S_{i}(v)% \quad\text{and}\quad 0\preceq T_{i}(u)\preceq T_{i}(v).$

In addition, $0\in D(L_{i})$ and $L_{i}(0)=0$ .
(H₃)

There exists $\psi_{i}\in K_{0}^{i}\setminus\{0\}$ such that for every $u\in K_{0}^{i}\cap X$ , one has

$u\preceq\left\|u\right\|\psi_{i}.$
(H₄)

There exists $\varphi_{i}\in K_{0}^{i}\setminus\{0\}$ such that for every $u\in K_{0}^{i}\cap X$ with $L_{i}u\in K_{0}^{i}$ , one has

$\left\|u\right\|\varphi_{i}\preceq u\quad(\text{abstract Harnack inequality}).$
(H₅)

$F_{i}:X^{2}\rightarrow 2^{Y}$ is usc, maps bounded sets into bounded sets and

$F_{i}((K_{0}^{1}\times K_{0}^{2})\cap X^{2})\subset K_{0}^{i}.$
(H₆)

For each vector $\lambda=(\lambda_{1},\lambda_{2})\in(0,+\infty)^{2}$ , there exist elements $\underline{F_{i}}(\lambda),\overline{F_{i}}(\lambda)\in K_{0}^{i}$ such that

$\underline{F_{i}}(\lambda)\preceq y\preceq\overline{F_{i}}(\lambda)\ \text{ % for every }y\in F_{i}(u_{1},u_{2})$

and every $(u_{1},u_{2})$ such that $u_{j}\in K_{0}^{j}\cap X$ and $\lambda_{j}\,\varphi_{j}\preceq u_{j}\preceq\lambda_{j}\,\psi_{j}$ , $j=1,2$ .

For each $i\in\{1,2\}$ , we consider the cone in $X$

K_{i}=\{u\in K_{0}^{i}\cap X\,:\,\left\|u\right\|\varphi_{i}\preceq u\}

and the product cone in $X^{2},$ $K=K_{1}\times K_{2}$ .

Let $N_{i}:X^{2}\rightarrow 2^{X}$ be the operators

(3.3)

N_{i}(u_{1},u_{2})=L_{i}^{-1}F_{i}(u_{1},u_{2}),\quad i=1,2.

Note that we can write $N=(N_{1},N_{2})=(\Psi_{1}\Phi_{1},\Psi_{2}\Phi_{2})$ , where

(3.4)

\Psi_{i}=T_{i},\quad\Phi_{i}=S_{i}\,F_{i}\quad\text{for }i=1,2.

For $r,R\in(0,+\infty)^{2}$ , $r=\left(r_{1},r_{2}\right),\ R=\left(R_{1},R_{2}\right),$ the notation $K_{r,R}$ stands for

K_{r,R}=\{u\in K\,:\,r_{i}\leq\left\|u_{i}\right\|\leq R_{i}\ \text{ for }i=1,2\},

and from (H₃) and (H₄), for every $(u_{1},u_{2})\in K_{r,R}$ one has

r_{j}\,\varphi_{j}\preceq\left\|u_{j}\right\|\varphi_{j}\preceq u_{j}\preceq% \left\|u_{j}\right\|\psi_{j}\preceq R_{j}\psi_{j},\quad j=1,2.

Observe that clearly for $i=1,2$ , the map $\Psi_{i}$ is usc with convex and compact values since it is a single-valued continuous map. Moreover, $\Phi_{i}$ is usc (as the composition of a continuous single-valued map and an usc multivalued map) and $\Phi_{i}(K_{r,R})$ is relatively compact since $F_{i}(K_{r,R})$ is bounded and $S_{i}$ maps bounded sets into relatively compact sets.

Lemma 3.1.

Assume that conditions (H₁)-(H₅) hold. Then for $N=(N_{1},N_{2}),$ one has

N(K)\subset K.

Proof.

Let us see that $N_{i}(K)\subset K_{i}$ for $i=1,2$ . Indeed, if $u\in K$ and $v\in N_{i}(u)=L_{i}^{-1}F_{i}(u),$ then $u_{1}\in K_{0}^{1}\cap X$ , $u_{2}\in K_{0}^{2}\cap X$ and thus, by (H₅), $F_{i}(u)\subset K_{0}^{i}$ . Next, by (H₂), we obtain that $v\in L_{i}^{-1}F_{i}(u)\subset K_{0}^{i}\cap D(L_{i})\subset K_{0}^{i}\cap X$ . Finally, $L_{i}v\in F_{i}(u)$ and $F_{i}(u)\subset K_{0}^{i}$ imply that $\left\|v\right\|\varphi_{i}\preceq v$ , as a consequence of the abstract Harnack inequality in (H₄). Therefore, $v\in K_{i}$ , as wished. ∎

Now we present the following general existence and localization result.

Theorem 3.2.

Assume that for each $i\in\{1,2\}$ , conditions (H₁)-(H₆) hold and there exist $\alpha_{i},\beta_{i}>0$ with $\alpha_{i}\neq\beta_{i}$ such that

(3.5)		$\displaystyle L_{i}^{-1}\underline{F_{i}}(\left\\|u_{1}\right\\|,\left\\|u_{2}% \right\\|)$	$\displaystyle\nprec u_{i}\quad\text{for all }u\in K_{r,R}\text{ with }\left\\|u_{i}\right\\|=\alpha_{i},$
(3.6)		$\displaystyle u_{i}$	$\displaystyle\nprec L_{i}^{-1}\overline{F_{i}}(\left\\|u_{1}\right\\|,\left\\|u_{% 2}\right\\|)\quad\text{for all }u\in K_{r,R}\text{ with }\left\\|u_{i}\right\\|=% \beta_{i},$

where $r_{i}=\min\{\alpha_{i},\beta_{i}\}$ and $R_{i}=\max\{\alpha_{i},\beta_{i}\}$ .

Then problem (3.1) has at least one positive solution $u\in K$ with

r_{i}\leq\left\|u_{i}\right\|\leq R_{i}\qquad(i=1,2).

Proof.

Let us apply Corollary 2.5 to the operator $N$ defined in (3.3) and the cone $K$ considered above. Note that $N$ is a decomposable map with $\Phi_{i}$ and $\Psi_{i}$ as in (3.4).

First, we prove that

(3.7)

L_{i}^{-1}y\nprec u_{i}\quad\text{for all }y\in F_{i}(u)\text{ and all }u\in K% _{r,R}\text{ with }\left\|u_{i}\right\|=\alpha_{i}.

By condition (H₆), $\underline{F_{i}}(\left\|u_{1}\right\|,\left\|u_{2}\right\|)\preceq y$ for all $y\in F_{i}(u_{1},u_{2})$ and all $u\in K_{r,R}$ . Hence, by (H₂), $L_{i}^{-1}\underline{F_{i}}(\left\|u_{1}\right\|,\left\|u_{2}\right\|)\preceq L% _{i}^{-1}y$ for all $y\in F_{i}(u_{1},u_{2})$ and all $u\in K_{r,R}$ . Now condition (3.5) implies (3.7).

Secondly, we show that

(3.8)

u_{i}\nprec L_{i}^{-1}y\quad\text{for all }y\in F_{i}(u)\text{ and all }u\in K% _{r,R}\text{ with }\left\|u_{i}\right\|=\beta_{i}.

By conditions (H₂) and (H₆), $L_{i}^{-1}y\preceq L_{i}^{-1}\overline{F_{i}}(\left\|u_{1}\right\|,\left\|u_{2% }\right\|)$ for all $y\in F_{i}(u_{1},u_{2})$ , $u\in K_{r,R}$ . Then condition (3.6) implies (3.8).

Therefore, Corollary 2.5 ensures the existence of at least one fixed point for $N$ located in $K_{r,R}$ , which is a positive solution of the system (3.1). ∎

It is said that the norm $\left\|\cdot\right\|$ of the Banach space $X$ is monotone with respect to the ordering given by the cones $K_{0}^{i}$ ( $i=1,2$ ) if $0\preceq x\preceq y$ implies $\left\|x\right\|\leq\left\|y\right\|$ . In that case, conditions (3.5) and (3.6) hold provided that (3.9) and (3.10) below are satisfied.

Corollary 3.3.

Assume that, for each $i\in\{1,2\}$ , conditions (H₁)-(H₆) hold, the norm $\left\|\cdot\right\|$ is monotone with respect to the ordering and there exist $\alpha_{i},\beta_{i}>0$ with $\alpha_{i}\neq\beta_{i}$ such that

(3.9)		$\displaystyle\left\\|L_{i}^{-1}\underline{F_{i}}(\left\\|u_{1}\right\\|,\left\\|u_% {2}\right\\|)\right\\|$	$\displaystyle>\alpha_{i}\quad\text{for all }u\in K_{r,R}\text{ with }\left\\|u_% {i}\right\\|=\alpha_{i},$
(3.10)		$\displaystyle\left\\|L_{i}^{-1}\overline{F_{i}}(\left\\|u_{1}\right\\|,\left\\|u_{% 2}\right\\|)\right\\|$	$\displaystyle<\beta_{i}\quad\text{for all }u\in K_{r,R}\text{ with }\left\\|u_{% i}\right\\|=\beta_{i},$

where $r_{i}=\min\{\alpha_{i},\beta_{i}\}$ and $R_{i}=\max\{\alpha_{i},\beta_{i}\}$ .

Then problem (3.1) has at least one positive solution $u\in K$ with

r_{i}\leq\left\|u_{i}\right\|\leq R_{i}\qquad(i=1,2).

A number of particular examples of differential problems which illustrate the applicability of the previous theory can be seen in [9] (see also [22]). Roughly speaking, the abstract theory works provided that a Harnack type inequality can be obtained for the problem.

4. Application to $\phi$ -Laplacian systems of inclusions

The aim of this section is to derive sufficient conditions for the existence and localization of positive solutions for systems of the form

(4.1)

\left\{\begin{array}[]{l}-\left(\phi_{1}(u_{1}^{\prime})\right)^{\prime}\in G_% {1}(t,u_{1},u_{2})\ \ \text{ in }(0,1)\\ -\left(\phi_{2}(u_{2}^{\prime})\right)^{\prime}\in G_{2}(t,u_{1},u_{2})\ \ % \text{ in }(0,1)\\ u_{1}^{\prime}(0)=u_{1}(1)=0=u_{2}^{\prime}(0)=u_{2}(1),\end{array}\right.

where for each $i\in\{1,2\}$ , $\phi_{i}:(-a_{i},a_{i})\rightarrow(-b_{i},b_{i})$ ( $0<a_{i},b_{i}\leq+\infty$ ) is an increasing homeomorphism such that $\phi_{i}(0)=0$ and $G_{i}:(0,1)\times\mathbb{R}_{+}^{2}\rightarrow 2^{\mathbb{R}_{+}}$ is an usc multivalued map with closed convex values, which maps bounded sets into bounded sets.

Problem (4.1) can be studied by means of the abstract scheme described in Section 3. Here $X=C[0,1]$ , $Y=L^{\infty}(0,1)$ , $\left\|\cdot\right\|=\left\|\cdot\right\|_{\infty}$ the sup-norm in $X$ , $P$ is the positive cone of $C[0,1]$ , $K_{0}^{i}$ is the positive cone of $L^{\infty}(0,1)$ , $\psi_{i}\equiv 1$ and

L_{i}(w)(t)=-\left(\phi_{i}(w^{\prime}(t))\right)^{\prime},\quad F_{i}(u_{1},u% _{2})=G_{i}(\cdot,u_{1}(\cdot),u_{2}(\cdot))\qquad(i=1,2).

Note that, for each $w\in L^{\infty}(0,1)$ and each $i\in\{1,2\}$ , one has

(4.2)

L_{i}^{-1}(w)(t)=-\int_{t}^{1}\phi_{i}^{-1}\left(-\int_{0}^{s}w(\tau)\,d\tau% \right)ds

and $\ L_{i}^{-1}=T_{i}\circ S_{i},$ with

T_{i}(w)(t)=-\int_{t}^{1}\phi_{i}^{-1}\left(-w(s)\right)ds,\qquad S_{i}(w)(t)=% \int_{0}^{t}w(\tau)\,d\tau.

In the sequel, we assume that $G_{i}$ is such that $S_{i}\circ F_{i}$ is an usc multivalued map with closed and convex values. Clearly, conditions (H₁), (H₂), (H₃) and (H₅) hold.

On the other hand, condition (H₄) holds here thanks to the following Harnack type inequality proved in [10] for the differential operator $Lu:=-\left(\phi(u^{\prime})\right)^{\prime}$ (where $\phi:(-a,a)\rightarrow(-b,b)$ is an increasing homeomorphism with $\phi(0)=0$ ) and the boundary conditions $u^{\prime}(0)=u(1)=0.$

Proposition 4.1.

For each $c\in(0,1)$ and any $u\in C^{1}[0,1]$ with $u^{\prime}(0)=u(1)=0,$ $u^{\prime}(t)\in(-a,a)$ for every $t\in[0,1],$ $\phi\circ u^{\prime}\in W^{1,1}(0,1)$ and $\left(\phi(u^{\prime})\right)^{\prime}\leq 0$ on $[0,1],$ the following inequality holds:

u(t)\geq(1-c)\left\|u\right\|_{\infty}\quad\text{for all \ }t\in[0,c].

Hence, condition (H₄) holds by taking the function $\varphi_{i}$ given by

\varphi_{i}(t)=\left\{\begin{array}[]{ll}1-c_{i},&\text{ for }t\in[0,c_{i}]\\ 0,&\text{ for }t\in(c_{i},1],\end{array}\right.

where $0<c_{i}<1$ is fixed.

Assume that for each $i\in\{1,2\},$ the multivalued map $G_{i}$ satisfies the following condition:

(C)

There exist continuous functions $\ f_{i},h_{i}:[0,1]\times\mathbb{R}_{+}^{2}\rightarrow\mathbb{R}_{+}$ such that
1. (i)
  
  $f_{i}$ and $h_{i}$ are nondecreasing in the second and third variables on $\mathbb{R}_{+}$ ;
2. (ii)
  
  for every $(t,x,y)\in[0,1]\times\mathbb{R}_{+}^{2}$ ,
  
  $f_{i}(t,x,y)\leq z\leq h_{i}(t,x,y)<b_{i}\quad\text{for all }z\in G_{i}(t,x,y).$

Finally, hypothesis (H₆) is satisfied for

	$\displaystyle\underline{F_{i}}(\lambda_{1},\lambda_{2})(t)$	$\displaystyle=f_{i}(t,\varphi_{1}(t)\lambda_{1},\varphi_{2}(t)\lambda_{2}),$
	$\displaystyle\overline{F_{i}}(\lambda_{1},\lambda_{2})(t)$	$\displaystyle=h_{i}(t,\lambda_{1},\lambda_{2}).$

Therefore, as a consequence of Corollary 3.3, we obtain the following result concerning the existence of positive solutions for (4.1).

Theorem 4.2.

Assume that, for each $i\in\{1,2\}$ , condition $(C)$ holds and there exist $\alpha_{i},\beta_{i}>0$ with $\alpha_{i}\neq\beta_{i}$ such that

(4.3)		$\displaystyle-\int_{c_{1}}^{1}\phi_{1}^{-1}\left(-\int_{0}^{c_{1}}f_{1}(\tau,(% 1-c_{1})\alpha_{1},\varphi_{2}(\tau)r_{2})\,d\tau\right)ds>\alpha_{1},$
(4.4)		$\displaystyle-\int_{c_{2}}^{1}\phi_{2}^{-1}\left(-\int_{0}^{c_{2}}f_{2}(\tau,% \varphi_{1}(\tau)r_{1},(1-c_{2})\alpha_{2})\,d\tau\right)ds>\alpha_{2},$
(4.5)		$\displaystyle-\int_{0}^{1}\phi_{1}^{-1}\left(-\int_{0}^{s}h_{1}(\tau,\beta_{1}% ,R_{2})\,d\tau\right)ds<\beta_{1},$
(4.6)		$\displaystyle-\int_{0}^{1}\phi_{2}^{-1}\left(-\int_{0}^{s}h_{2}(\tau,R_{1},% \beta_{2})\,d\tau\right)ds<\beta_{2},$

where $r_{i}=\min\{\alpha_{i},\beta_{i}\}$ and $R_{i}=\max\{\alpha_{i},\beta_{i}\}$ .

Then problem (4.1) has at least one positive solution $u=(u_{1},u_{2})\in K$ with

r_{i}\leq\left\|u_{i}\right\|_{\infty}\leq R_{i}\qquad(i=1,2).

Proof.

Let us apply Corollary 3.3 with the operators $L_{i}^{-1}$ as defined in (4.2).

First, let us show that condition (3.9) is satisfied for $i=1$ (the case $i=2$ is analogous). Since for each nonnegative function $w\in L^{\infty}(0,1)$ ,

\left\|L_{1}^{-1}(w)\right\|_{\infty}=-\int_{0}^{1}\phi_{1}^{-1}\left(-\int_{0% }^{s}w(\tau)\,d\tau\right)ds,

by the monotonicity assumptions on $f_{1}$ , we have that for each $u\in K_{r,R}$ with $\left\|u_{1}\right\|_{\infty}=\alpha_{1},$

	$\displaystyle\left\\|L_{1}^{-1}\underline{F_{1}}(\left\\|u_{1}\right\\|_{\infty},% \left\\|u_{2}\right\\|_{\infty})\right\\|_{\infty}$	$\displaystyle\geq-\int_{0}^{1}\phi_{1}^{-1}\left(-\int_{0}^{s}f_{1}(\tau,% \varphi_{1}(\tau)\alpha_{1},\varphi_{2}(\tau)r_{2})\,d\tau\right)ds$
		$\displaystyle\geq-\int_{c_{1}}^{1}\phi_{1}^{-1}\left(-\int_{0}^{s}f_{1}(\tau,% \varphi_{1}(\tau)\alpha_{1},\varphi_{2}(\tau)r_{2})\,d\tau\right)ds$
		$\displaystyle\geq-\int_{c_{1}}^{1}\phi_{1}^{-1}\left(-\int_{0}^{c_{1}}f_{1}(% \tau,(1-c_{1})\alpha_{1},\varphi_{2}(\tau)r_{2})\,d\tau\right)ds.$

Hence, applying (4.3), we deduce that condition (3.9) in Corollary 3.3 holds.

Finally, we check that condition (3.10) is also satisfied for $i=2$ (the case $i=1$ is similar). We have that for each $u\in K_{r,R}$ with $\left\|u_{2}\right\|_{\infty}=\beta_{2},$

\left\|L_{2}^{-1}\overline{F_{2}}(\left\|u_{1}\right\|_{\infty},\left\|u_{2}% \right\|_{\infty})\right\|\leq-\int_{0}^{1}\phi_{2}^{-1}\left(-\int_{0}^{s}h_{% 2}(\tau,R_{1},\beta_{2})\,d\tau\right)ds,

and so the conclusion follows from (4.6). ∎

Note that, in particular, if $\ c_{1}=c_{2}=:c$ and $\phi_{1}$ and $\phi_{2}$ are odd homeomorphisms, then conditions (4.3)-(4.6) hold if the following inequalities are satisfied:

	$\displaystyle(1-c)\phi_{1}^{-1}\left(c\min_{\tau\in[0,c]}f_{1}(\tau,(1-c)% \alpha_{1},(1-c)r_{2})\right)$	$\displaystyle>\alpha_{1},$
	$\displaystyle(1-c)\phi_{2}^{-1}\left(c\min_{\tau\in[0,c]}f_{2}(\tau,(1-c)r_{1}% ,(1-c)\alpha_{2})\right)$	$\displaystyle>\alpha_{2},$
	$\displaystyle\phi_{1}^{-1}\left(\max_{\tau\in[0,1]}h_{1}(\tau,\beta_{1},R_{2})\right)$	$\displaystyle<\beta_{1},$
	$\displaystyle\phi_{2}^{-1}\left(\max_{\tau\in[0,1]}h_{2}(\tau,R_{1},\beta_{2})\right)$	$\displaystyle<\beta_{2}.$

Next we give an example of application of Theorem 4.2, where the operators associated to the two equations of the system have different behaviors: compression for one of them and expansion for the other one.

Example 4.3.

Consider the system

(4.7)

\left\{\begin{array}[]{l}-\left(\displaystyle\frac{u^{\prime}}{\sqrt{1+{u^{% \prime}}^{2}}}\right)^{\prime}\in G_{1}(u,v)\ \ \text{ in }(0,1)\\[14.22636pt] -v^{\prime\prime}\in G_{2}(u,v)\ \ \text{ in }(0,1)\\[5.69046pt] u^{\prime}(0)=u(1)=0=v^{\prime}(0)=v(1),\end{array}\right.

where the usc multivalued maps $G_{1}$ and $G_{2}$ are given by

(4.8)

G_{1}(u,v)=\left\{\begin{array}[]{ll}\frac{1}{2}\left([\sqrt[3]{u},\sqrt[4]{u}% ]e^{-u}+\cos^{2}v\right)&\quad\text{if }u\in[0,1]\\[5.69046pt] \frac{1}{2}\left(e^{-1}+\cos^{2}v\right)&\quad\text{if }u>1\end{array}\right.

and

G_{2}(u,v)=[1+\sin^{2}u,2]v^{2}.

One may easily verify that condition (C) holds for the functions $f_{i}$ and $h_{i}$ ( $i=1,2$ ) defined as

f_{1}(u)=\left\{\begin{array}[]{ll}\min\left\{\frac{1}{2}\sqrt[3]{u}e^{-u},% \frac{1}{2e}\right\}&\quad\text{if }u\in[0,1],\\[5.69046pt] \frac{1}{2e}&\quad\text{if }u>1,\end{array}\right.

h_{1}(u)=4/5,\ \ \ f_{2}(v)=v^{2},\ \ \ h_{2}(v)=2v^{2}.

Moreover, choosing $c=1/2$ , straightforward computations show that we can take $\alpha_{1}=1/50,$ $\beta_{1}=2,$ $\alpha_{2}=18$ and $\beta_{2}=1/3.$ Therefore, according to Theorem 4.2, problem (4.7) has at least one positive solution $(u,v)$ such that

\frac{1}{50}\leq\left\|u\right\|_{\infty}\leq 2\quad\text{and}\quad\frac{1}{3}% \leq\left\|v\right\|_{\infty}\leq 18.

Remark 4.1 (Asymptotic conditions).

As shown in Example 4.3, it is meaningful the simple case where condition (C) is given by functions of the form $f_{i}(t,u_{1},u_{2})=f_{i}(u_{i})$ and $h_{i}(t,u_{1},u_{2})=h_{i}(u_{i})$ for $i=1,2.$

In this case, the existence of the numbers $\alpha_{i}$ is guaranteed by the following asymptotic behavior at zero or infinity:

\limsup\limits_{\lambda\rightarrow 0^{+}}\frac{(1-c)\phi_{i}^{-1}\left(cf_{i}(% (1-c)\lambda)\right)}{\lambda}>1\quad\text{or}\quad\limsup\limits_{\lambda% \rightarrow+\infty}\frac{(1-c)\phi_{i}^{-1}\left(cf_{i}((1-c)\lambda)\right)}{% \lambda}>1.

Similarly, the existence of the numbers $\beta_{i}$ can be obtained from the following asymptotic behavior at zero or infinity:

\liminf\limits_{\lambda\rightarrow 0^{+}}\frac{\phi_{i}^{-1}\left(h_{i}(% \lambda)\right)}{\lambda}<1\quad\text{or}\quad\liminf\limits_{\lambda% \rightarrow+\infty}\frac{\phi_{i}^{-1}\left(h_{i}(\lambda)\right)}{\lambda}<1.

Remark 4.2 (Multiple solutions).

Multiplicity results can be immediately established if several pairs of numbers $(\alpha_{1},\beta_{1})$ or $(\alpha_{2},\beta_{2})$ as in (4.3)-(4.6) exist. Note that we may obtain multiple solutions with multiplicity not necessarily concerned with all components of the solution, as shown in the following example.

Example 4.4.

Consider the system (4.7) with $G_{1}$ as defined in (4.8) and

G_{2}(u,v)=[1+\sin^{2}u,2]v^{2}+\frac{1}{9}\sqrt[3]{v}.

To check condition (C), take $f_{1}$ and $h_{1}$ as in Example 4.3,

f_{2}(v)=v^{2}+\frac{1}{9}\sqrt[3]{v}\quad\text{and}\quad h_{2}(v)=2v^{2}+% \frac{1}{9}\sqrt[3]{v}.

Again, with $c=1/2,$ one may easily verify that conditions (4.3)-(4.6) hold by taking $\alpha_{1}=1/50,$ $\beta_{1}=2$ and as pair $\left(\alpha_{2},\beta_{2}\right),$ any one of the following two pairs $\left(1/500,1/4\right),$ $\left(20,1/3\right).$

Thus Theorem 4.2 applied twice ensures the existence of at least two positive solutions $(u_{1},v_{1})$ and $(u_{2},v_{2})$ such that

\frac{1}{50}\leq\left\|u_{1}\right\|_{\infty},\left\|u_{2}\right\|_{\infty}% \leq 2,\quad\frac{1}{500}\leq\left\|v_{1}\right\|_{\infty}\leq\frac{1}{4}\quad% \text{and}\quad\frac{1}{3}\leq\left\|v_{2}\right\|_{\infty}\leq 20.

Acknowledgements

Jorge Rodríguez-López was partially supported by Xunta de Galicia ED431C 2019/02.

References

[1] C. Bereanu, P. Jebelean, J. Mawhin, Non-homogeneous boundary value problems for ordinary and partial differential equations involving singular $\phi$ -Laplacians, Mat. Contemp. 36 (2009), 51–65.
[2] A. Cabada and R. L. Pouso, Existence results for the problem $\left(\phi\left(u^{\prime}\right)\right)^{\prime}=f\left(t,u,u^{\prime}\right)$ with nonlinear boundary conditions, Nonlinear Anal. 35 (1999), no. 2, 221–231.
[3] J. Á. Cid and P.J. Torres, Solvability for some boundary value problems with $\phi$ -Laplacian operators, Discrete Cont. Dynamic. Syst. A 23 (2009), 727–732.
[4] I. Coelho, C. Corsato, F. Obersnel and P. Omari, Positive solutions of the Dirichlet problem for the one-dimensional Minkowski-curvature equation, Adv. Nonlinear Stud. 12 (2012), 621–638.
[5] J.-F. Couchouron and R. Precup, Homotopy method for positive solutions of $p$ -Laplace inclusions, Topol. Methods Nonlinear Anal. 30 (2007), 157–169.
[6] L. Ferracuti and F. Papalini, Boundary-value problems for strongly nonlinear multivalued equations involving different $\phi$ -Laplacians, Adv. Differential Equations 14 (2009), 541–566.
[7] Z. He and L. Miao, Uniqueness and multiplicity of positive solutions for one-dimensional prescribed mean curvature equation in Minkowski space, AIMS Mathematics 5 (2020) (4): 3840–3850.
[8] J. Henderson and H. Wang, An eigenvalue problem for quasilinear systems, Rocky Mountain J. Math. 37 (2007), no. 1, 215–228.
[9] D.-R. Herlea, D. O’Regan and R. Precup, Harnack type inequalities and multiple solutions in cones of nonlinear problems, Z. Anal. Anwend. 39 (2020), 151–170.
[10] D.-R. Herlea and R. Precup, Existence, localization and multiplicity of positive solutions to $\phi$ -Laplace equations and systems, Taiwan. J. Math. 20 (2016), 77–89.
[11] P. Jebelean and R. Precup, Symmetric positive solutions to a singular $\phi$ -Laplace equation, J. London Math. Soc. 99 (2019), 495–515.
[12] P.S. Kelevedjiev and S.A. Tersian, The barrier strip technique for a boundary value problem with $p$ -Laplacian, Electron. J. Differential Equations 28 (2013), 1–8.
[13] R. López Pouso, R. Precup and J. Rodríguez-López, Positive solutions for discontinuous systems via a multivalued vector version of Krasnosel’skiĭ’s fixed point theorem in cones, Mathematics 7:451 (2019), 1–15.
[14] R.Y. Ma and Y.Q. Lu, Multiplicity of positive solutions for second order nonlinear Dirichlet problem with one-dimension Minkowski-curvature operator, Adv. Nonlinear Stud. 15 (2015), 789–803.
[15] R.Y. Ma, L. Zhang and R. Liu, Existence results for nonlinear problems with $\phi$ -Laplacian, Electron. J. Qual. Theory Differ. Equ. 2015, No. 22, 1–7.
[16] J. Mawhin, Resonance problems for some non-autonomous ordinary differential equations, in Stability and Bifurcation Theory for Non-Autonomous Differential Equations, 103–184, Lectures Notes in Math., 2065, Springer, Heidelberg, 2013.
[17] D. O’Regan and R. Precup, Positive solutions of nonlinear systems with $p$ -Laplacian on finite and semi-infinite intervals, Positivity 11 (2007), 537–548.
[18] M. Pei, L. Wang and X. Lv, Existence and multiplicity of positive solutions of a one-dimensional mean curvature equation in Minkowski space, Bound. Value Probl. (2018) 2018: 43, 1–11.
[19] R. Precup, Fixed point theorems for decomposable multi-valued maps and applications, Z. Anal. Anwend. 22 (2003), 843–861.
[20] R. Precup, A vector version of Krasnosel’skiĭ’s fixed point theorem in cones and positive periodic solutions of nonlinear systems, J. Fixed Point Theory Appl. 2 (2007), 141–151.
[21] R. Precup, Componentwise compression-expansion conditions for systems of nonlinear operator equations and applications, Mathematical models in engineering, biology and medicine, 284–293, AIP Conf. Proc., 1124, Amer. Inst. Phys., Melville, NY (2009).
[22] R. Precup and J. Rodríguez-López, Multiplicity results for operator systems via fixed point index, Results Math. 74 (2019), 25: 1–14.
[23] R. Precup and J. Rodríguez-López, Fixed point index theory for decomposable multivalued maps and applications to discontinuous $\phi$ -Laplacian problems, Nonlinear Anal. 199 (2020), 111958.
[24] Z. Yang, Positive solutions for a system of $p$ -Laplacian boundary value problems, Comput. Math. Appl. 62 (2011), 4429–4438.

[1] A.N. Alahmadi, M. Alkan, and S. López-Permouth, Poor modules: the opposite of injectivity, Glasg. Math. J. 52(A) (2010), 7–17. doi: 10.1017/S001708951000025X [Crossref] [Web of Science ®], [Google Scholar]

[2] F.W. Anderson and K.R. Fuller, Rings and Categories of Modules, 2nd. Ed., Graduate Texts in Mathematics, Vol. 13, Springer-Verlag, New York, 1992. [Crossref], [Google Scholar]

[3] P. Aydogdu and B. Saraç, On Artinian rings with restricted class of injectivity domains, J. Algebra 377 (2013), 49–65. doi: 10.1016/j.jalgebra.2012.11.027 [Crossref] [Web of Science ®], [Google Scholar]

[4] G. Baccella, Semi-Artinian V-rings and semi-Artinian von Neumann regular rings, J. Algebra 173(3) (1995), 587–612. doi: 10.1006/jabr.1995.1104 [Crossref] [Web of Science ®], [Google Scholar]

[5] N. Er, S. López-Permouth, and N. Sokmez, Rings Whose Modules have Maximal or Minimal Injectivity Domains, J. Algebra 330(1) (2011), 404–417. doi: 10.1016/j.jalgebra.2010.10.038 [Crossref] [Web of Science ®], [Google Scholar]

[6] L. Gillman and M. Jerison, Rings of continuous functions. The University Series in Higher Mathematics, D. Van Nostrand Co., Inc., Princeton, N.J./Toronto/London/New York, 1960. [Crossref], [Google Scholar]

[7] R.P. Kurshan, Rings whose cyclic modules have finitely generated socles, J. Algebra 15 (1970), 376–386. doi: 10.1016/0021-8693(70)90066-9 [Crossref] [Web of Science ®], [Google Scholar]

[8] L.S. Levy and J.C. Robson, Hereditary Noetherian prime rings and idealizers, Mathematical Surveys and Monographs, Vol. 174, American Mathematical Society, Providence, RI, 2011. [Crossref], [Google Scholar]

[9] S. López-Permouth and J.E. Simental, Characterizing rings in terms of the extent of the injectivity and projectivity of their modules, J. Algebra 362 (2012), 56–69. doi: 10.1016/j.jalgebra.2012.04.005 [Crossref] [Web of Science ®], [Google Scholar]

[10] S. Mohammad and B. Müller, Continuous and Discrete Modules, London Mathematical Society Lecture Note Series, Vol. 147, Cambridge University Press, Cambridge, 1990. [Crossref], [Google Scholar]

[11] M.A. Mulero, Algebraic properties of rings of continuous functions, Fund. Math. 149(1) (1996), 55–66. doi: 10.4064/fm-149-1-55-66 [Crossref] [Web of Science ®], [Google Scholar]

[12] B.H. Neumann, On ordered division rings, Trans. Amer. Math. Soc. 66(1) (1949), 202–252. doi: 10.1090/S0002-9947-1949-0032593-5 [Crossref], [Google Scholar]

[13] S.S. Page and M.F. Yousif, Relative injectivity and chain conditions, Comm. Algebra 17(4) (1989), 899–924. doi: 10.1080/00927878908823766 [Taylor & Francis Online] [Web of Science ®], [Google Scholar]

[14] R. Wisbauer, Foundations of module and ring theory. A handbook for study and research, Gordon and Breach Science Publishers, Philadelphia, PA, 1991. [Google Scholar]

2022

Stationary Kirchhoff equations and systems with reaction terms

June 23, 2022

On high-order conforming finite element methods for ill-posed Helmholtz problems

January 7, 2022

On some applications of the controllability principle for fixed point equations

August 17, 2022

(3.9)		$\displaystyle\left\\|L_{i}^{-1}\underline{F_{i}}(\left\\|u_{1}\right\\|,\left\\|u_% {2}\right\\|)\right\\|$	$\displaystyle>\alpha_{i}\quad\text{for all }u\in K_{r,R}\text{ with }\left\\|u_% {i}\right\\|=\alpha_{i},$
(3.10)		$\displaystyle\left\\|L_{i}^{-1}\overline{F_{i}}(\left\\|u_{1}\right\\|,\left\\|u_{% 2}\right\\|)\right\\|$	$\displaystyle<\beta_{i}\quad\text{for all }u\in K_{r,R}\text{ with }\left\\|u_{% i}\right\\|=\beta_{i},$

Componentwise localization of solutions to systems of operator inclusions via Harnack type inequalities

Abstract

Authors

Keywords

Paper coordinates

PDF

About this paper

Journal

Publisher Name

DOI

Print ISSN

Online ISSN

Paper (preprint) in HTML form

Componentwise localization of solutions to systems of operator inclusions via Harnack type inequalities

Abstract.

1. Introduction

2. Existence results for operator systems with decomposable multivalued maps

2.1. A Schauder type fixed point theorem

Theorem 2.1.

Proof.

Corollary 2.2.

Proof.

2.2. A vector version of Krasnosel’skiĭ’s fixed point theorem in cones

Theorem 2.3.

Proof.

Theorem 2.4.

Corollary 2.5.

3. Positive solutions for systems of operator inclusions

Lemma 3.1.

Proof.

Theorem 3.2.

Proof.

Corollary 3.3.

4. Application to ϕ-Laplacian systems of inclusions

Proposition 4.1.

Theorem 4.2.

Proof.

Example 4.3.

Remark 4.1 (Asymptotic conditions).

Remark 4.2 (Multiple solutions).

Example 4.4.

Acknowledgements

References

References

Related Posts

4. Application to $\phi$ -Laplacian systems of inclusions